The Clarifai API's v2 design introduces various conventions aimed at simplifying the integration of AI functionalities into applications while maintaining a consistent and extensible user experience. The API supports both HTTP and gRPC protocols, with a REST architecture that conceals the complexity of underlying machine learning algorithms. It employs standard HTTP methods, such as POST for creating new objects, GET for retrieving objects, PATCH for modifying existing objects, and DELETE for removing objects. These methods are designed to handle multiple objects in bulk and include clear naming and structural conventions, such as using snake_case for JSON fields and ensuring request and response object types are interchangeable. The API's architecture allows for object-oriented operation through gRPC protobuffers, enhancing ease of use and flexibility. The documentation encourages users to include user_id and app_id in URLs for improved collaboration, and emphasizes that endpoints and object names should be plural to handle lists of items. These design choices are informed by lessons learned from Clarifai's v1 API, aiming to provide a more robust experience for users.